It is now evident that the source of the new 44 and 88Hz dark fringe noise is the WE CD fans tray.
A question is "which is the path"? Possibilities are (1) magnetic fields force on WE mirror magnets, and (2) pick-up of EM noise by the CD cables to the WE mirror coils. I remind that in 2009 we had similar 44Hz and 88Hz peaks from the CD crate at NE. The noise path was demonstrated to be EM-pick-up and it was succesfully mitigated by modifing the fan tray, i.e. distanciating a bit the fans from the CD electronics (see Suppression of 44Hz and 88Hz bumps). The present WE CD crate is already of this modified type. As can be read in the plot of that eLog (calibrating by the proper hrec TF) the noise was then reduced from hrec=5e-22 to hrec<=2e-22 (at 44Hz) and from hrec=7e-22 to hrec<=7e-23 (at 88Hz). You can see in 29318 that the presently observed hrec noise is compatible with these upper limits.
On the other hand, the magnetic path seems very unlikely. I made a rough computation of the expected noise using the the magnetic TF measured at WE with old payloads (Projection of far-field magnetic noise.) Which reads: TF*f^2 = 5e-10 at 44Hz, and 1e-10 at 88Hz (units of h*Hz^2/Tesla). The fans are at a similar distance (~3-4m) from the magnetic probes (Em_MABDWE*) as they are from the WE mirror. Thus assuming the probes reading at 44 and 88Hz give a measure of the magnetic field at the mirror (this ignoring the ~20Hz-pole low-pass magnetic filter of the oven and vac chamber, thus again overestimating the effect), I get for the magnetic filelds: 5e-11T at 44Hz and 3e-11T at 88Hz. And the predicted hrec noise would be: 1.5e-23 at 44 Hz and 4e-25 at 88Hz. This is a factor 20 lower than the observed noise at 44Hz and a factor 250 lower than the observed noise at 88Hz.
A mitigation of the noise requires to further displace ( a bit more than doubling) the fans from the CD boards. Might be not obvious.